This issue has probably been discussed before but I've been thinking about it for the last few weeks so I'll give it a go anyway.

There may be an issue with not knowing the total bitcoin supply that is currently available due to lost keys and from other factors. Would a solution to this issue be to destroy, or rather make coins that hasn't moved in a certain time period locked from further usage?

Since there are no way to tell the difference between coins that are locked due to lost keys and coins that hasn't moved for a time period, there are uncertainty to the total bitcoin supply that currently exists. This could be a potential problem if big amounts of coins are put back into circulation after being stale for a longer period.

Take for example the Satoshi addresses, which, afaik, are assumed to contain some 500 000 to one million coins. We will never be able to tell if these are taken out of circulation or just kept for usage further down the road, unless they start to move. This creates a big uncertainty and could potentially cause major effects if they start to be used after ten years or more. The same can of course be said about any of the top 100 addresses or any addresses with "stale" coins. If they where to be remained untouched for say 50 or 100 years and then suddenly start to move, the perceived available bitcoin supply would suddenly grow, potentially causing instability within the system.

Would a solution be to build into the protocol that coins that haven't moved for a given period, for example 20 years, become locked and thereby taken out of circulation? This would make it much easier to estimate the current coin supply and thereby making it easier to valuate a single bitcoin. This would also force entities to reveal their plan for addresses with large balances, point in case are the Satoshi addresses just mentioned.

Is this a problem or should we just always assume that there are (will be) 21 million coins available, disregarding how long coins have been sitting on a specific address?